view mercurial/stack.py @ 38513:6db38c9d7e00

revlog: efficient implementation of 'descendant'

Iterating over descendants is costly, because there are no "parent -> children"
pointers. Walking the other way around is much more efficient, especially on
large repositories, where descendant walks can cost seconds. And the other hand,
common ancestors code follows links in the right direction and has a compiled
implementation.

In real life usage, this saved up to 80s during some pull operations, where
descendant test happens in extension code.

# stack.py - Mercurial functions for stack definition
#
#  Copyright Matt Mackall <[e-mail]> and other
#
# This software may be used and distributed according to the terms of the
# GNU General Public License version 2 or any later version.

from __future__ import absolute_import

from . import (
    revsetlang,
    scmutil,
)

def getstack(repo, rev=None):
    """return a sorted smartrev of the stack containing either rev if it is
    not None or the current working directory parent.

    The stack will always contain all drafts changesets which are ancestors to
    the revision and are not merges.
    """
    if rev is None:
        rev = '.'

    revspec = 'reverse(only(%s) and not public() and not ::merge())'
    revset = revsetlang.formatspec(revspec, rev)
    revisions = scmutil.revrange(repo, [revset])
    revisions.sort()
    return revisions